BRENNEN

Industries / Frontier science / Longevity

Drugs

This work looks for compounds that slow aging itself, and tests them first in animals, usually by feeding them to mice for life and counting how long the mice live.

7

Milestones achieved, from 2009 to 2024.

3

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 2009

    Rapamycin extends mouse life

    The National Institute on Aging's Interventions Testing Program (ITP) runs every test at three sites, the Jackson Laboratory, the University of Michigan and UT Health San Antonio, in UM-HET3 mice, which are bred from four strains so that each mouse differs genetically and no result hangs on one strain or one lab. Rapamycin, a drug used to stop the rejection of transplanted organs, raised both median lifespan (the age by which half have died) and maximum lifespan in both sexes, even when first given at 600 days of age; measured at the age when 90% had died, the gain was 14% for females and 9% for males. The authors wrote that, to their knowledge, these were the first results to show a drug extending lifespan in both sexes of a mammal.

    Lessona drug first given late in life can still lengthen it.

    pmc.ncbi.nlm.nih.govnia.nih.gov

    Credit

    D. E. Harrison, R. Strong, R. A. Miller and colleagues, ITP

  2. 2013

    A sirtuin company closes

    GlaxoSmithKline (GSK) bought Sirtris Pharmaceuticals of Cambridge, Massachusetts, in 2008 for US$720 million. Sirtris claimed that compounds such as resveratrol, a molecule found in red wine, directly activate SIRT1, an enzyme tied to aging, and GSK's own scientists could not confirm it. In 2011 the ITP reported that resveratrol, at two doses, did not change survival in male or female mice. In March 2013 GSK closed Sirtris and moved its programs into its own research.

    Lessona claim about how a compound works, and a compound that fails an outside lifespan test, are weak grounds for a drug company.

    blogs.nature.compmc.ncbi.nlm.nih.gov

    Credit

    Sirtris and GSK; R. A. Miller and colleagues, ITP

  3. 2014

    Some drugs favor males

    In the ITP, acarbose, a diabetes drug that slows the breakdown of carbohydrate in the gut, raised male median lifespan by 22% but female median lifespan by only 5%. 17-alpha-estradiol, a "non-feminizing" form of estrogen, raised male median lifespan by 12% and did not change female lifespan, and NDGA, a plant compound, helped males only, even at a dose that gave females similar blood levels. Differences in body weight did not explain the gaps.

    Lessoneach sex has to be tested and reported on its own, because a drug can work in one and not the other.

    pmc.ncbi.nlm.nih.gov

    Credit

    D. E. Harrison, R. A. Miller and colleagues, ITP

  4. 2015

    Drugs that clear senescent cells

    Senescent cells have stopped dividing but do not die, and release signals that inflame the tissue around them. A team including Y. Zhu, T. Tchkonia and J. L. Kirkland found that these cells depend on survival networks, and named drugs that kill them selectively "senolytics". The cancer drug dasatinib together with quercetin, a plant compound, reduced senescent cells in old mice and improved heart and blood-vessel function five days after one dose. In 2018 the same group reported that giving the pair intermittently to old mice raised survival after treatment by 36%.

    pmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.gov

    Credit

    Y. Zhu, T. Tchkonia, J. L. Kirkland and colleagues

  5. 2020

    Canagliflozin helps male mice

    Canagliflozin is a diabetes drug that blocks the kidneys from taking glucose back into the blood. Given from 7 months of age in the ITP, it raised male median survival by 14%, with similar effects at each of the three sites, and gave no lifespan gain to females. The authors noted that acarbose, which blunts glucose surges in a different way, also helps mainly males, and pointed to daily glucose peaks as a target.

    Lessontwo drugs with different mechanisms but one shared effect, lower glucose peaks, gave the same male-only pattern.

    pmc.ncbi.nlm.nih.gov

    Credit

    R. A. Miller, R. Strong and colleagues, ITP

  6. 2024

    NR and fisetin fail

    Nicotinamide riboside (NR) is a form of vitamin B3 that cells turn into NAD, a molecule they use to make energy; fisetin is a plant compound tested as a senolytic. In 2021 the ITP reported that NR did not extend lifespan in either sex, "despite some data suggesting that nicotinamide riboside would be effective". In 2024 it reported the same for fisetin, given daily or in cycles from 20 months of age, while astaxanthin and meclizine raised male median lifespan by 12% and 8%.

    Lessonmost compounds fail; of the 99 finished ITP test arms the NIA listed in April 2026, 65 showed no lifespan gain in either sex, and each failure holds only for the dose and schedule used.

    pmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.govnia.nih.gov

    Credit

    D. E. Harrison, R. Strong, R. A. Miller and colleagues, ITP

  7. 2024

    The 900-day rule

    K. Pabis, B. K. Kennedy and colleagues reanalyzed published mouse lifespan studies and found that control groups that die young make a treatment look more effective than it is. They proposed that, without an independent repeat, a result in well-studied inbred or UM-HET3 mice deserves high confidence only when the control mice reach a median lifespan close to 900 days, or when the treated mice live well past it.

    Lessonhow long the control mice lived matters as much as the percentage gain.

    ebi.ac.uk

    Credit

    K. Pabis, D. Barardo, J. Gruber, M. Kaeberlein, B. K. Kennedy and colleagues, in Ageing Research Reviews

Watch for

  1. Trametinib plus rapamycin

    Trametinib is a cancer drug that blocks the Ras-MEK-ERK growth signal inside cells. L. Partridge's department at the Max Planck Institute for Biology of Ageing in Cologne reported that trametinib alone raised median lifespan by 7.2% in female and 10.2% in male mice, and that trametinib with rapamycin raised it by 29% in females and 27% in males, with fewer liver tumors. The mice were one hybrid strain, tested at one site.

    Achieved whenan independent test reports a significant lifespan gain, for example the ITP arm that has fed trametinib to UM-HET3 mice from 12 months of age since its 2024 cohort, listed as in progress.

    pmc.ncbi.nlm.nih.govnia.nih.gov

    Claimed by

    L. Gkioni, L. Partridge and colleagues, in Nature Aging, May 2025

  2. A lifespan pill for dogs

    Loyal, a San Francisco company, says the US Food and Drug Administration's Center for Veterinary Medicine has accepted two of the three main technical sections of its application for conditional approval of LOY-002, a daily pill meant to extend the healthy lifespan of senior dogs: effectiveness in February 2025 and target animal safety on 13 January 2026. It says its STAY study enrolled 1,300 dogs at 70 veterinary clinics by July 2025, and that manufacturing is the section still to come.

    Achieved whenthe FDA grants LOY-002 conditional approval, which by Loyal's account would be the first FDA approval of a drug for lifespan extension in any species.

    loyal.com

    Claimed by

    Loyal, January 2026

  3. Semaglutide and aging pace

    Semaglutide is a GLP-1 drug, one of a class used for diabetes and obesity. A team led from UC San Diego went back to a finished trial of 108 adults with HIV and excess abdominal fat, about half given semaglutide and the rest a placebo (a dummy treatment), and reported that the drug slowed DunedinPACE, an epigenetic clock that estimates the pace of aging from chemical marks on DNA, by 9%. The trial was not designed to measure aging, and the lead author said "We are not saying that semaglutide reverses aging".

    Achieved whena randomized trial that names an aging measure as its main endpoint in advance, in people without HIV and run by another group, finds a similar slowing.

    today.ucsd.eduebi.ac.uk

    Claimed by

    Michael Corley and colleagues, in Nature Communications, May 2026

Open

  1. 01

    Forecast the mouse tests

    A public, time-stamped set of forecasts, from AI models and from people, of whether each ITP test arm still running will significantly extend median lifespan in male mice and in female mice, scored against the results as the ITP publishes them.

    The brief
    Credit · unclaimed
  2. 02

    AI curation of lifespan papers

    A tool that reads a published mouse lifespan study and fills in a structured record of it: compound, dose in parts per million of food, route, sex, strain, age at first dose, control median lifespan, percentage change and significance. Its records are scored field by field against records curated by hand.

    The brief
    Credit · unclaimed

Open 01 · the brief

Forecast the mouse tests

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public, time-stamped set of forecasts, from AI models and from people, of whether each ITP test arm still running will significantly extend median lifespan in male mice and in female mice, scored against the results as the ITP publishes them.

Why it is still open

The NIA's ITP table, last updated on 13 April 2026, lists 19 arms in progress from the 2021 to 2025 cohorts, among them taurine, urolithin A, trametinib, sildenafil, captopril and a rapamycin-estradiol mix. Prediction markets on Manifold, set up by Kamil Pabis in February 2022, covered only the 2017 cohort and drew 4 to 10 traders each. A. V. Belikov and colleagues trained classifiers on the DrugAge database to predict lifespan-extending compounds and tested picks in worms, but scored them on past data, not on live ITP arms.

nia.nih.govmanifold.marketsebi.ac.uk

Built with what exists

The ITP results table and published papers as a base rate (34 of 99 finished arms showed a gain in at least one sex); DrugAge as background; language models that return a probability for each fixed-answer question, such as Gemini Decision 1 (see the Note "A thousand questions, one pass"); an Open Science Framework registration to fix the forecasts in time.

  1. First step

    Copy the ITP table into a spreadsheet, compute the base rate by sex and by drug class, and register a first forecast sheet this month.

  2. Done looks like

    Forecasts for all 19 running arms, by sex, registered before any of their results appear, from at least three AI models and at least ten people, with a Brier score (the average squared error of a probability forecast) published for each forecaster as each result arrives.

Open 02 · the brief

AI curation of lifespan papers

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A tool that reads a published mouse lifespan study and fills in a structured record of it: compound, dose in parts per million of food, route, sex, strain, age at first dose, control median lifespan, percentage change and significance. Its records are scored field by field against records curated by hand.

Why it is still open

DrugAge, the main database of compounds tested for lifespan, says it "is manually curated by experts", and its Build 6 of 15 November 2026 holds 3,441 entries. Its 2025 update re-recorded mouse data from the original papers by hand and added control lifespans, which the 900-day rule depends on. In May 2026 Arnas Stučinskas and Audronė Jakaitienė of Vilnius University described a pipeline in which language models pulled 2,987 graded claims from 1,797 papers on aging in people; it covers human studies, not mouse lifespan tests, and was not scored against DrugAge. No published test of AI extraction against DrugAge records turned up in a search on 19 November 2026.

genomics.senescence.infopmc.ncbi.nlm.nih.govjournals.vu.lt

Built with what exists

Current language models with document reading, the Europe PMC full-text service for open-access papers, and DrugAge Build 6 mouse records as the answer key.

  1. First step

    Pick 50 mouse entries in DrugAge Build 6 whose papers are open access, run one model on them, and count how often it gets the control lifespan right.

  2. Done looks like

    Records extracted from at least 200 mouse lifespan papers, with accuracy reported for each field against DrugAge, and every disagreement listed with the page it came from, released under an open license.